<p>Recently, Content-based image retrieval (CBIR) systems are emerging as a promising solution for assisting medical practitioners to efficiently access the extensive image databases. However, the CBIR systems face difficulty in research fields due to the massive growth of multimedia content on the internet. Further, the existing techniques fail to capture the latent characteristics associated with the images and obtain limited accuracy, especially for medical images. Hence, this research proposes the Military Collie Optimization based Support Vector Machine enabled Deep Convolutional Neural Network (MC-SDCN) for achieving an effective CBIR system. Specifically, the proposed MC-SDCN approach combines the synergic strength of both the CNN and SVM, resulting in capturing the intricate patterns and latent characteristics of medical images for retrieving the most relevant images. Specifically, the Radial Basis Function (RBF) in SVM employs the distance measure as the similarity metric to retrieve the most similar images that exist in the indexed database. Besides, the Military Collie Optimization (MC) is exploited for fine-tuning the hyperparameters of the MC-SDCN model. Experimental results demonstrate that the proposed MC-SDCN for CBIR achieves the high accuracy of 93.31%, precision of 94.31%, recall of 93.98%, and F1-score of 94.15% with 80 number of retrievals, surpassing the other existing techniques.</p>